CN102057679B - Compressed image noise removal device and reproduction device - Google Patents
Compressed image noise removal device and reproduction device Download PDFInfo
- Publication number
- CN102057679B CN102057679B CN2009801208521A CN200980120852A CN102057679B CN 102057679 B CN102057679 B CN 102057679B CN 2009801208521 A CN2009801208521 A CN 2009801208521A CN 200980120852 A CN200980120852 A CN 200980120852A CN 102057679 B CN102057679 B CN 102057679B
- Authority
- CN
- China
- Prior art keywords
- image
- noise remove
- section
- information
- noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013139 quantization Methods 0.000 claims abstract description 86
- 238000013519 translation Methods 0.000 claims description 25
- 238000007906 compression Methods 0.000 claims description 21
- 230000006835 compression Effects 0.000 claims description 21
- 238000006243 chemical reaction Methods 0.000 claims description 18
- 238000012937 correction Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000000034 method Methods 0.000 description 19
- 239000000203 mixture Substances 0.000 description 16
- 238000011002 quantification Methods 0.000 description 15
- 230000009467 reduction Effects 0.000 description 14
- 230000002093 peripheral effect Effects 0.000 description 12
- 238000001914 filtration Methods 0.000 description 11
- 238000011156 evaluation Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 9
- 230000008030 elimination Effects 0.000 description 8
- 238000003379 elimination reaction Methods 0.000 description 8
- 238000007689 inspection Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000008034 disappearance Effects 0.000 description 4
- 241000255925 Diptera Species 0.000 description 3
- 238000013144 data compression Methods 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002715 modification method Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 101000969688 Homo sapiens Macrophage-expressed gene 1 protein Proteins 0.000 description 1
- 102100021285 Macrophage-expressed gene 1 protein Human genes 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000002266 amputation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 239000007799 cork Substances 0.000 description 1
- 230000003467 diminishing effect Effects 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000000518 rheometry Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
- H04N19/86—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression involving reduction of coding artifacts, e.g. of blockiness
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/117—Filters, e.g. for pre-processing or post-processing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/136—Incoming video signal characteristics or properties
- H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/18—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a set of transform coefficients
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/44—Decoders specially adapted therefor, e.g. video decoders which are asymmetric with respect to the encoder
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/80—Details of filtering operations specially adapted for video compression, e.g. for pixel interpolation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/48—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using compressed domain processing techniques other than decoding, e.g. modification of transform coefficients, variable length coding [VLC] data or run-length data
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
- H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
- H04N19/51—Motion estimation or motion compensation
- H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A compressed image noise removal device includes: a decoder means for decoding a digital image-compressed stream; an information holding unit for holding a plurality of blocks of sub information decoded by a VLD unit; a noise judgment unit which performs a noise removal judgment on a display image generated by the decoder means and the information holding unit; a noise removal unit for executing a noise removal on the block judged to be subjected to a noise removal by the noise judgment unit, by using image data outputted from an inverse quantization unit, motion compensation data outputted from a motion compensation unit, and sub information held in the information holding unit; and a display image holding unit which holds as a display image, the output image of the noise removal unit ifthe noise judgment unit has judged to execute a noise removal or the output of the decoder means if the noise judgment unit has judged not to execute a noise removal.
Description
The application advocates the preference of the Japanese patent application that proposed on July 30th, 2008 2008-195634 number, by with reference to its content is introduced among the application.
Technical field
This patent is take the digitized video machine of the reproduction of carrying out digital compressed image as object.
Background technology
As the background technology of the art, Japanese kokai publication hei 6-70173 (patent documentation 1) is for example arranged.Put down in writing in this communique that " [purpose] prevents the generation of vibrating in the reproduced image.The state of the image that [structure] obtains according to the coded image of synthetic each piece determines whether with edge detection unit 22 to have the edge.For marginate, carry out plane approximation with the zone of the both sides of plane approximation section 32 edges, generate approximate image.Then this approximate image is carried out DCT, transform to frequency domain.Afterwards, it is supplied to operational part 38, with the interior proportion by subtraction addition of decoded picture to stipulate of frequency domain.That is, in operational part, add the high-frequency region of losing because of quantification treatment.Because the disappearance of this high-frequency region is the reason that produces vibration, by the processing in the arithmetic unit, can prevent the vibration in the reproduced image.”
Patent documentation 1: Japanese kokai publication hei 6-70173
Summary of the invention
In recent years, digital broadcasting is popularized gradually, also can enjoy like a cork the image of high definition at home.Because this digital picture has more amount of information, usually with the international standard mode MPEG2 of dynamic image compression, H.264 wait and compress.In this compression, by image information is carried out frequency translation, and utilize and quantize to come the amputation radio-frequency component, carry out the reduction of data volume.Owing to being somebody's turn to do the reduction by the data volume that quantizes to carry out, in compressed image, can producing because of the radio-frequency component disappearance the fuzzy and mosquito noise of image.
For the removal of these noises, the method (Japanese kokai publication hei 6-70173) that the frequency content of being cut down is predicted has been proposed.But in this patent, only the image information with the object piece generates approximate image, and additional its radio-frequency component carries out noise-cut thus.Can imagine, the difference according to the precision of this approximate image may become the image different from the original image (original image) that carries out image compression, causes correction image different from all edge images, and the quality of reproduced image reduces.
Particularly, in the control of the quantization amplitude (quantization width) in the common dynamic image compression, visual characteristic based on the mankind, set quantization amplitude wider for the marginal portion, so the precision of this approximate image reduces more, can imagine that the quality that obtains final correction image also can reduce.
When digital compressed image was reproduced, hope can be predicted the composition that loses because of quantification accurately, obtained high-quality reproduced image.The object of the invention is to high-quality ground reproducing compressed image.
In order to reach above-mentioned purpose, adopt the structure of putting down in writing in the claim.
According to the present invention, can high-quality ground reproducing compressed image.
Problem, structure and effect beyond above-mentioned can be clear and definite by the explanation of following execution mode.
Description of drawings
Fig. 1 is the figure of an aspect of the basic conception of the present invention among expression the first embodiment.
Fig. 2 is the figure of the structure of expression the second embodiment.
Fig. 3 is the figure of the structure of expression the 3rd embodiment.
Fig. 4 is the figure of the use of the peripheral blocks of data among expression the second embodiment.
Fig. 5 is the figure of each signal component of the signal to be assumed to be picture signal of expression among the second embodiment when carrying out frequency translation.
Fig. 6 is the figure of the situation that among expression the second embodiment signal component shown in Figure 5 quantized.
Fig. 7 is the figure of structure of the situation of the use periphery blocks of data of expression among the second embodiment.
Fig. 8 is the figure of the structure of the use original image of expression among the 3rd embodiment noise remove of estimating computing.
Fig. 9 is the figure of the object lesson of the tabular of the coefficient of expression among the 3rd embodiment and coefficient correction.
Figure 10 is the figure of expression the 4th embodiment.
Figure 11 is the figure of expression the 5th embodiment.
Embodiment
Below use the description of drawings embodiments of the present invention.
Embodiment 1
Use description of drawings the first embodiment.The present embodiment is used for illustrating an aspect of basic conception of the present invention.Fig. 1 is the figure of an aspect of expression basic conception of the present invention.The former coefficient 101 of (1) expression of Fig. 1, the coefficient that expression has certain composition (value).The example that (2) expression of Fig. 1 quantizes with certain quantization amplitude 102 former coefficient.The so-called quantification refers to discrete values such as the continuous quantities such as analogue data being approximately numerical data.After former coefficient was quantized with certain quantization amplitude, the composition of former coefficient was similar to by quantization amplitude, became to quantize rear coefficient 107.
When digital data compression, directly do not use former coefficient yet, but by using this quantization amplitude and quantizing rear coefficient, realize that the compression of big data quantity is processed.
Then, the method for reducing that uses the numerical data after this quantification is compressed is described.When reduction, come reduction coefficient 105 is reduced with coefficient after quantization amplitude and the quantification.This processing is also referred to as inverse quantization sometimes.This processing is the example that value is reduced to intermediate value in the zone of quantization amplitude.Value is reduced to which position of this quantization amplitude, in quantizing side and inverse quantization side, defines in advance.
This composition that has reduced namely reduces the difference of value of the former coefficient before coefficient and the reduction and is called quantization error 104.This is that the factor digital data is compressed and the noise of generation.About the relation of this quantization amplitude and compression ratio, generally speaking, the data compression ratio uprises but quantization error becomes large when increasing quantization amplitude, during the lower quantization amplitude quantization error diminish but data compression also than diminishing.
The above object lesson of processing is described.For example, consider situation about when ultimate constituent has 55 this value, quantizing with quantization amplitude 20.In this situation, what use in the compression is the such information of coefficient 2 after quantizing, quantization amplitude 20.When carrying out inverse quantization based on this information, in the situation of the median that value is reduced to quantization amplitude, the reduction coefficient is 50, has produced the quantization error of (former coefficient) 55-(reduction coefficient) 50=5.
Removal about this quantization error, utilize the character of peripheral information and composition thereof to carry out the removal of noise for the reduction coefficient, but in the situation of the composition of revising the reduction coefficient in order to remove noise, shown in 106, need in the employed quantization amplitude of inverse quantization zone, revise composition.This is because if surpass the employed quantization amplitude of the inverse quantization shown in 106 zone because the processing that is used for the removal quantization error makes the composition of reduction coefficient, mean that then it becomes the coefficient (image) different from former coefficient (original image).Thus, verily carry out the reduction of former coefficient in order to carry out noise remove, the noise remove that carry out the noise remove of this principle, namely is no more than in the scope in quantization amplitude zone is important.
The present embodiment is to use MPEG2 (ISO/IEC 13813-2) mode as the example of image compression mode.But the present invention is not limited to the MPEG2 mode, also can enough other compress modes (such as H.264, MPEG1, MPEG4, JPEG, VC1) etc. realization.
The summary of the present embodiment at first is described.When image compression (coder side), image information is broken down into frequency content, and quantizes.The quantization amplitude of this moment is kept in the stream as the quantization amplitude of DC in quantized value, quantization matrix and the frame.At decoder-side, because can obtain coefficient after the quantification and the information of quantization amplitude, so this coefficient is positioned at the scope of which value in the time of can inferring coding when quantizing.When the decoding side is carried out noise remove, for the coefficient behind the inverse quantization, in the mode in the scope that drops on this quantization amplitude value is revised.
Each signal component (503～510) when Fig. 5 represents that the signal that is assumed to be picture signal carried out frequency translation.Among this Fig. 5, represent to be transformed to the example of 8 compositions of tieing up in order to simplify.Among the MPEG2, use and to carry out the DCT that is called of conversion by every 8x8 (Discrete Cosine Transform: frequency translation discrete cosine transform) transforms to 64 dimensions as this conversion.Transverse axis 501 expression frequency, to the right the direction indication radio-frequency components of Fig. 5.The value of the longitudinal axis 502 each frequency content of expression of Fig. 5, this frequency content of the larger expression of absolute value exists manyly.Then, in image compression, this frequency content is carried out quantification treatment.
Fig. 6 is the situation that expression quantizes signal component shown in Figure 5.Quantification is that each signal is calculated respectively quantization amplitude 601, quantizes by each composition.Quantization amplitude 601 is by being the quantized value that macro block (macroblock) is given by each processing unit of encoding, and the quantization matrix of giving by every width of cloth image decides.Generally speaking, preferably the radio-frequency component quantization amplitude is just larger in the image compression.With this quantization amplitude each composition is quantized respectively, coefficient and quantization amplitude form stream after will quantizing, and realize thus image compression.
About the decoding of this stream (reduction, decompress(ion)), reduce frequency content with quantization parameter and quantization amplitude, and this frequency content is carried out anti-frequency translation, be reduced to thus image information.Wherein, quantize necessary information, in the MPEG2 standard, indicated by DC quantization amplitude (intra_dc_precision) and Intra/Inter handover information (macroblock_intra) in quantized value (quantizer_scale), quantization matrix (non_intra/intra_quantizer_matrix), the frame.In other standards, being used for quantizing necessary information is object.
In the image, can produce noise because of the quantization error such as explanation among the embodiment 1 behind the front image of this image compression and the image restoring.This noise is commonly referred to as mosquito noise and block noise.About the removal of these noises, can be by applying various filtering and realize going back original image.But, in the correction of going back original image of carrying out noise remove, will not become the image different from the image before the compression if in the scope of the employed quantization amplitude of compression, do not revise.Removal and the front image of verily reduction compression in order to carry out noise need to carry out noise remove according to this principle.
The structure of expression the present embodiment in Fig. 2.The left half of Fig. 2 represents the decoder element 201 of MPEG2.The Video of MPEG2 (video) stream is input to VLD, and (Variable Length Decoding: variable-length decoding) module 202, utilize the variable length Gray code rheology to be changed into the movable informations 206 such as the quantitative information 205 that quantizes rear coefficient 208 and quantization amplitude 209 for calculating, and motion vector.Quantize rear coefficient by inverse quantization section 203 inverse quantizations, and in anti-frequency conversion part, carry out anti-frequency translation.Quantitative information 205 as quantization amplitude outputs to inverse quantization section and information storing section from VLD section.Movable information 206 is output to dynamic compensating unit 207.In dynamic compensating unit, according to movable information from before decoded picture generation forecast image, with the information addition of anti-frequency conversion part, generate to be used for thus the decoded picture 210 that shows.For this decoded picture, for as the using with reference to image of next image, in dynamic compensating unit, preserve.It more than is the explanation of decoder element.This is the action identical with the decoder of common MPEG2.Should based on the image of MPEG2 decoding, comprise mosquito noise and block noise as quantizing noise, so will remove with the module on right side.This noise remove is undertaken by the piece of the 8x8 of the unit of every DCT that processes as mpeg decode.
In addition, there is unobvious this feature of noise in the violent part of image motion in the dynamic image.Therefore, in the violent part of moving, also can be judged as and not carry out noise remove.
In addition, also with good grounds user's hobby and have the situation of the pattern of not carrying out noise remove.Also can specify to be judged as according to this user and not carry out noise remove.
Describe for frequency conversion part 252.In this module, carry out frequency translation for the image that has carried out noise remove.As the mode of frequency translation, adopt the mode corresponding with anti-frequency translation.Use DCT in the situation of MPEG.
Describe for quantization amplitude inspection section 253.In this module, check according to coefficient after quantizing and quantization amplitude, judge whether the image behind the noise remove is present in the interior zone of quantization amplitude.
Check by this quantization amplitude, under all coefficients all drop on situation in the quantization amplitude, judge that then it is suitable revising, with this image as showing that image is sent to the Image Saving section 254 that shows.In the inspection of quantification amplitude, be checked through in the situation that outside the quantization amplitude, can again be sent to noise remove section with this information and carry out noise remove, also can be with the image before the noise remove as showing that image be sent to the Image Saving section that shows.
The use of peripheral blocks of data then, is described with Fig. 4 A, 4B.The decoded picture 401 that Fig. 4 A, 4B represent to decode and finish.Decoding is carried out take the macro block 402 of rectangle as unit.Quantized value also can be by this macro block unit's set point.Fig. 4 A is the scene of having taken background 404 and object 403, and quantizing noise 405 comprises more generation in the macro block of object and background at the same time.The example of the stride (body at quarter) 406 of the quantized value 407 of this moment represents in Fig. 4 B.This is the quantized value of setting when coding, in general, according to the mankind's visual characteristic, more carefully quantizes in more smooth part, and the part at the edge more slightly quantizes.In such situation, the information of peripheral piece that is used for carrying out the noise remove of certain piece is utilized, and is used to the information from the less piece of quantized value, and the possibility that can obtain effective information is higher.Can think that this is that the disappearance of high-frequency data is less because of in the less piece of this quantized value, the quality of reproduced image is also better.In addition, among Fig. 4 A, for the piece that comprises object and background, use background information 408, object to use the mode of object information 409 to carry out noise remove with background, can carry out more efficiently noise remove.Particularly, the information during as the predicted picture of generted noise removal section is used with this peripheral information.In addition, except the method for this use peripheral information, the information of the information of the frame in use past and future, use brightness and aberration etc. also is effective.
Among Fig. 7, expression is used for using the structure of this periphery piece.Structure at Fig. 2 newly increases memory section 701, preserves thus quantized value and decoded picture, uses in noise remove section.
Also can in this memory section, preserve the frame behind the noise remove and be used in noise remove after the next frame.In this situation, use movable information to carry out motion compensation for the image behind the noise remove, then carry out noise remove as predicted picture.
In addition, also can in memory section, preserve quantized value, this quantized value is used in noise remove.This is because varying in size of quantizing in I-image and the P/B-image so by using the quantized value in the I-image to carry out continuously noise remove, can process equably the noise remove of each image, can realize the high-quality noise remove under the dynamic image.
Embodiment 3
The noise remove of the second embodiment, whether the image after employing is implemented noise remove and checked this noise remove decoded picture drops on the structure in the quantization amplitude, but among the 3rd embodiment, describe for the embodiment that revises in frequency content in the decode procedure.
The structure of expression the present embodiment among Fig. 3.Decoder element 201 among the figure is identical with the second embodiment.
Describe for frequency conversion part 301.Frequency conversion part carries out frequency translation to the predicted picture from dynamic compensating unit output.About the method for frequency translation, adopt the method corresponding with anti-frequency conversion part 204.Be DCT in the situation of MPEG.
Then, carry out the addition 302 of the output of frequency conversion part and inverse quantization section 203.Thus, generate with decoded picture is carried out DCT after identical information.301 and 302 operation is the operation be used to the frequency translation information that obtains decoded picture, also can adopt the mode of the decoded picture that is generated by decoder element being implemented DCT.
Describe revising amplitude calculating part 303.Utilize decoder element to obtain the information of the rear coefficient of quantification and quantization amplitude, calculate the correction amplitude of the frequency content in the zone that is no more than quantization amplitude for the frequency content of decoded picture.
Showing Image Saving section 254, is to preserve the module that shows image.
In this module, also can carry out anti-frequency translation, carry out noise at image-region and judge.In addition, also can judge with the piece of peripheral piece, past and future.But, in this situation, need to preserve the information that will judge with memory.
Carry out the explanation of noise remove section 304.Noise remove section is to carrying out noise remove by the information of the frequency content of the decoded picture of 302 adder output.The correction amplitude of the value of the noise remove under this frequency content is revised in the amplitude that is calculated by the amplitude of correction calculating part.At this moment, modification method can be considered following methods.
Make the composition medelling of the original images such as natural image, adopt the pattern close with the frequency content of coded image.
Method shown in Figure 8 (hereinafter narration)
Method for the former is described in detail.This is the method for carrying out noise remove in frequency content.Natural image or artificial image generally are the less clearly images of noise.If these images are carried out frequency translation with the conversion of DCT equifrequent, can bring characteristic transformation results.To carry out after the frequency translation coefficient table of gained to various patterns such as natural image, artificial images in advance formats.Then, from form, search for immediate pattern for the coefficient after quantizing, revise the coefficient after quantizing.
Among Fig. 9, the tabular of expression coefficient and the object lesson of coefficient correction.901 expression frequency transform coefficients in video codec, imagination is used the situation of the DCT of 8x8 in this example, so there are 64 coefficients.904 (the DC compositions) take upper left coefficient as flip-flop, right represents the transverse direction composition from the low-frequency component to the radio-frequency component, lower direction represents that the longitudinal direction composition is from the low-frequency component to the radio-frequency component.When tabular, natural image and artificial image are implemented DCT, be the combination of low-frequency component group 902 (the oblique line piece among the figure) by the group that is present in low frequency, the component list of radio-frequency component group 903 (the pattern-free piece among the figure) is formatted.This is because for quantization parameter, mostly quantification can cause radio-frequency component by more reduction in the situation, and information more keeps in the low-frequency component, so preferred such structure.In addition, the division of this low-frequency component group and radio-frequency component group can be determined by the memory space of the form that can guarantee.Can guarantee in the situation of more form, need to enlarge the low frequency region group, otherwise it is dwindled.Make that the low frequency region class range is larger can more correctly to carry out the correction of coefficient.
Then, the modification method of coefficient after the quantification of using this form described.Low-frequency component group coefficient by the coefficient behind the inverse quantization in when decoding extracts the radio-frequency component group with reference to above table, and the value of the radio-frequency component group of the coefficient behind the inverse quantization is revised.Particularly, be that it is effective coming value is replaced by the value of reference form in 0 the situation in the radio-frequency component group of the coefficient behind the inverse quantization.At this moment, preferably each radio-frequency component group is also carried out the affirmation of the quantization amplitude of explanation among the embodiment 1.
In addition, also can be from the frequency content of as shown in Figure 4 peripheral piece and the image of past and future with reference to the similar information of piece, carry out thus the correction of the coefficient behind the inverse quantization.
Then carry out the latter's explanation.Among Fig. 8, expression uses original image to estimate the structure of the noise remove of computing.On the structure of Fig. 3, newly add original image evaluation computing 802 modules.
Original image is estimated operational part, to the demonstration image by anti-frequency conversion part input, calculates evaluation of estimate according to evaluation function.This evaluation function is preferably and provides the evaluation function that has more distinctive numerical value than natural image and artificial image etc.In addition, when calculating evaluation function, also can not only use the information of processing object piece inside, also use the interior information of frame of peripheral piece and past and future to calculate.In this situation, use other memory to preserve and be used for carrying out the data that evaluation of estimate is calculated.
The use of the peripheral block message during then, to the generation forecast image describes.In the moving picture encoding, with the block unit switching quantized value of image inside.For example among the MPEG2, switch by the block unit with the 16x16 pixel, can with frame in the tendency of image as one man carry out image compression.Can think that the disappearance of high-frequency data is less in the less piece of this quantized value, the quality of reproduced image is also better.Thus, the adjacent block of the object piece of investigating in the situation that has the less piece of quantized value, is used the information of this piece energetically.Particularly, in predicted picture generated, the adjacent pixels of the adjacent block that quantized value is less was as the value of the predicted picture of this piece.By like this, also can expect to reduce the effect of the piece distortion that interblock produces.The quantitative information of this periphery piece is realized by information storing section.Its example of expression in Fig. 3.For decoded picture, in the situation of the stride of considering quantized value, utilize the view data of peripheral piece.
In the present embodiment, expression is the structure in the situation of (ITU-T H.264)/MPEG4AVC (ISO/IEC14496-10) and VC1 (the dynamic image compress mode that Microsoft develops is got by Windows (registered trade mark) Media Video 9 standardization) H.264.In the standard of these moving picture encodings, also carry out the noise remove based on the principle of embodiment 1.In addition, in the present embodiment, adopt the viewpoint roughly the same with the mode of embodiment 2, but because the coding standard that relates to is different, so coding tools is different, the direct mode of Application Example 2.Thus, especially describe the part different from embodiment 2 in detail.
In these standards, decoded picture is implemented to be called the filtering processing of filtering (block elimination filtering) in the ring.
The structure chart of expression the present embodiment among Figure 10.For the stream of Video, in VLD section 1002, carry out the decoding of Variable Length Code, information etc. that be divided into view data, be used for to calculate quantized value and matrix quantizes the sub-information such as quantitative information 1005, movable information 1006 and block mode information of amplitude.For view data, in inverse quantization section 1003, carry out inverse quantization and process, in anti-frequency conversion part 1004, carry out anti-frequency translation.At this moment, inverse quantization module is supplied with quantitative information.Wherein, H.264 with VC1 in, the cut size of piece is the variable-sized of 4x4～8x8.These information are also supplied with each module from the VLD module.
On the other hand, movable information 1006 supplies to dynamic compensating unit 1007 and carries out motion compensation.By with the data after this motion compensation and anti-frequency conversion part addition, generate decoded picture 1010.
In addition, in the standard of the present embodiment, the pattern of only using the information in the decoded frame to carry out predictive coding that is called the Intra prediction is arranged, but the processing of Intra prediction is undertaken by this dynamic compensating unit equally in the present embodiment.Then, this decoded picture in carrying out the DF section 1011 of block elimination filtering, carries out block noise and removes.These process the image that consists of common decoder H.264/VC1, as decoder element 1001.
Describe for noise judging part, noise remove section, quantization amplitude inspection section.This module has the function same with noise judging part, noise remove section, the quantization amplitude inspection section of Fig. 2 basically.But in the standard of the present embodiment, the frequency translation size is the variable-sized of 4x4～8x8, so correspondingly process with this frequency translation size.About the piece cut size, can obtain by the sub-information from the VLD module size of DCT and learn.
Describe for DF2 section.In this module, for being judged as the piece that does not need noise to judge in the image that has been carried out noise remove by noise remove section or the noise judging part, implement block elimination filtering.But, for the image that has been carried out noise remove by noise remove section, because carried out noise remove, so also can implement block elimination filtering without this module.
In addition, in the situation that carried out noise remove, also can switch according to its result and carry out or do not carry out block elimination filtering.
In addition, whether DF2 section is the module with function identical with DF section except in the situation that carried out noise remove with noise remove section and implement the judgement of block elimination filtering, can use identical circuit.
Carry out the structure of noise remove during the standards such as H.264 Figure 11 represents, VC1 at frequency domain.Decoder element 1101 is structures same with the decoder element 1001 of Figure 10.
About adder 1152, frequency conversion part 1151, noise judging part 1160, noise remove section 1155, revise amplitude calculating part 1153, anti-frequency conversion part 1154, show Image Saving section 1157, all have and the structure of Fig. 3 same function roughly.But in the standard of the present embodiment, the variable-block that the piece of DCT is of a size of 4x4～8x8 is cut apart, and quantizes also to carry out with this piece size, so will process under the prerequisite of recognizing this piece size.
Abovely with reference to accompanying drawing suitable execution mode of the present invention is illustrated, but the present invention is not limited to these examples certainly.For example, above-described embodiment is the detailed description of carrying out for the explanation the present invention of easy to understand ground, might not be defined in to comprise all illustrated structures.In the category of the technological thought that industry practitioner puts down in writing in the claims, can expect various modifications or fixed case, these also belong to technical scope of the present invention certainly.
Claims (6)
1. a compressed image noise remove device is characterized in that, comprising:
Decoder element, it comprises:
Stream after the digital image compression is decoded as the VLD section of view data and sub-information;
The view data that is got by the decoding of VLD section is carried out the inverse quantization section of inverse quantization;
Data behind the inverse quantization are carried out the anti-frequency conversion part of anti-frequency translation; With
Based on the dynamic compensating unit that the sub-information that is got by the decoding of VLD section is carried out motion compensation, information storing section, it preserves a plurality of sub-information that is got by the decoding of VLD section;
The noise judging part, it carries out the noise remove judgement to the demonstration image and the information storing section that are generated by decoder element;
Noise remove section, it is to being judged to be the piece that carries out noise remove by the noise judging part, use view data from the output of inverse quantization section, from the moving compensating data of dynamic compensating unit output, be kept at the sub-information the information storing section, carry out noise remove; With
Show Image Saving section, it is being judged to be by the noise judging part when carrying out noise remove, and the output image of noise remove section is saved as the demonstration image, is being judged to be when not carrying out noise remove, and the output of decoding unit is saved as the demonstration image,
Described noise remove is undertaken by inverse quantization section output factor is carried out the coefficient correction,
This revised coefficient value, with according to calculate from the sub-information of VLD section output quantization amplitude when again quantizing, carry out the correction in the scope of coefficient value of inverse quantization section output.
2. compressed image noise remove device as claimed in claim 1 is characterized in that:
The sub-information of using in the noise remove section is the information of the quantization amplitude (intra_dc_precision) of DC in the quantized value (quanter_scale) that uses in quantizing, quantization matrix (quantiser_matrix), the frame.
3. compressed image noise remove device as claimed in claim 1 is characterized in that:
Whether carry out the judgement of noise remove by marginal information or the quantized value decision of this piece.
4. compressed image noise remove device as claimed in claim 1 is characterized in that:
The correction value of the dequantized coefficients when carrying out noise remove uses the view data of the piece that quantized value is less in the adjacent block of this piece to calculate.
5. compressed image noise remove device as claimed in claim 1 is characterized in that:
Decoding unit is the MPEG2 as the international standard mode of image compression.
6. compressed image noise remove device as claimed in claim 1 is characterized in that:
As the international standard mode of image compression H.264 decoding unit is.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2008-195634 | 2008-07-30 | ||
| JP2008195634 | 2008-07-30 | ||
| PCT/JP2009/060422 WO2010013543A1 (en) | 2008-07-30 | 2009-06-08 | Compressed image noise removal device and reproduction device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102057679A CN102057679A (en) | 2011-05-11 |
| CN102057679B true CN102057679B (en) | 2013-10-23 |
Family
ID=41610249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009801208521A Active CN102057679B (en) | 2008-07-30 | 2009-06-08 | Compressed image noise removal device and reproduction device |
Country Status (8)
| Country | Link |
|---|---|
| US (1) | US8929439B2 (en) |
| EP (1) | EP2320662A4 (en) |
| JP (1) | JP5362723B2 (en) |
| KR (1) | KR101223780B1 (en) |
| CN (1) | CN102057679B (en) |
| BR (1) | BRPI0913589B1 (en) |
| MY (1) | MY154427A (en) |
| WO (1) | WO2010013543A1 (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2015122619A (en) * | 2013-12-24 | 2015-07-02 | キヤノン株式会社 | Information processing device, information processing method, and program |
| US9552628B2 (en) * | 2014-03-12 | 2017-01-24 | Megachips Corporation | Image processor |
| RU2646358C2 (en) * | 2016-04-07 | 2018-03-02 | Николай Дмитриевич Егоров | Method for eliminating redundancy when compressing sequence of images |
| DE102018112215B3 (en) * | 2018-04-30 | 2019-07-25 | Basler Ag | Quantizer determination, computer readable medium, and apparatus implementing at least two quantizers |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1334682A (en) * | 1996-05-28 | 2002-02-06 | 松下电器产业株式会社 | Image predictive coding/decoding device and method and recording medium |
| CN1835548A (en) * | 2005-03-18 | 2006-09-20 | 富士施乐株式会社 | Decoding apparatus, decoding method and program product therefor |
| US20070160129A1 (en) * | 2006-01-10 | 2007-07-12 | Tatsuro Fujisawa | Video decoding apparatus and video decoding method |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0670173A (en) | 1992-08-17 | 1994-03-11 | Kyocera Corp | Picture decoding device and picture coding/decoding system |
| JP3536355B2 (en) | 1994-06-09 | 2004-06-07 | ソニー株式会社 | Encoding device |
| TW293227B (en) | 1994-11-24 | 1996-12-11 | Victor Company Of Japan | |
| JP2940422B2 (en) * | 1994-11-24 | 1999-08-25 | 日本ビクター株式会社 | Method for reducing block distortion generated when decoding transform-coded image data and decoding device for transform-coded image data |
| US5852475A (en) * | 1995-06-06 | 1998-12-22 | Compression Labs, Inc. | Transform artifact reduction process |
| JP2005117680A (en) | 1995-12-28 | 2005-04-28 | Sony Corp | Image signal coding method and device, image signal transmission method and decoding method and device |
| US6256349B1 (en) * | 1995-12-28 | 2001-07-03 | Sony Corporation | Picture signal encoding method and apparatus, picture signal transmitting method, picture signal decoding method and apparatus and recording medium |
| KR100229796B1 (en) * | 1996-08-30 | 1999-11-15 | 전주범 | Image decoding system including compensation at detoriorate image |
| KR100269125B1 (en) * | 1997-10-25 | 2000-10-16 | 윤덕용 | Image post processing method and apparatus for reducing quantization effect |
| US20060093232A1 (en) * | 2002-12-10 | 2006-05-04 | Koninkjkle Phillips Electronics N.V | Unified metric for digital video processing (umdvp) |
| US20050243914A1 (en) * | 2004-04-29 | 2005-11-03 | Do-Kyoung Kwon | Adaptive de-blocking filtering apparatus and method for mpeg video decoder |
| JP5089188B2 (en) | 2007-02-09 | 2012-12-05 | キヤノン株式会社 | Amine compound, electrophotographic photosensitive member, process cartridge, and electrophotographic apparatus |
-
2009
- 2009-06-08 US US12/993,920 patent/US8929439B2/en active Active
- 2009-06-08 MY MYPI2010005634A patent/MY154427A/en unknown
- 2009-06-08 CN CN2009801208521A patent/CN102057679B/en active Active
- 2009-06-08 WO PCT/JP2009/060422 patent/WO2010013543A1/en active Application Filing
- 2009-06-08 KR KR1020107027122A patent/KR101223780B1/en active IP Right Grant
- 2009-06-08 EP EP09802788A patent/EP2320662A4/en not_active Withdrawn
- 2009-06-08 BR BRPI0913589-8A patent/BRPI0913589B1/en active IP Right Grant
- 2009-06-08 JP JP2010522654A patent/JP5362723B2/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1334682A (en) * | 1996-05-28 | 2002-02-06 | 松下电器产业株式会社 | Image predictive coding/decoding device and method and recording medium |
| CN1835548A (en) * | 2005-03-18 | 2006-09-20 | 富士施乐株式会社 | Decoding apparatus, decoding method and program product therefor |
| US20070160129A1 (en) * | 2006-01-10 | 2007-07-12 | Tatsuro Fujisawa | Video decoding apparatus and video decoding method |
Also Published As
| Publication number | Publication date |
|---|---|
| US8929439B2 (en) | 2015-01-06 |
| KR20100137022A (en) | 2010-12-29 |
| KR101223780B1 (en) | 2013-01-17 |
| CN102057679A (en) | 2011-05-11 |
| EP2320662A4 (en) | 2011-11-02 |
| EP2320662A1 (en) | 2011-05-11 |
| JPWO2010013543A1 (en) | 2012-01-12 |
| WO2010013543A1 (en) | 2010-02-04 |
| JP5362723B2 (en) | 2013-12-11 |
| BRPI0913589B1 (en) | 2020-11-24 |
| MY154427A (en) | 2015-06-15 |
| US20110064135A1 (en) | 2011-03-17 |
| BRPI0913589A2 (en) | 2015-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1513349B1 (en) | Bitstream-controlled post-processing video filtering | |
| CN105430398B (en) | Method and apparatus for in-loop de-artifact filtering | |
| JP5805097B2 (en) | Method and apparatus for DC intra prediction mode for video encoding and decoding | |
| EP2323407A1 (en) | Video image encoding method, video image decoding method, video image encoding apparatus, video image decoding apparatus, program and integrated circuit | |
| US20080192824A1 (en) | Video coding method and video coding apparatus | |
| EP2168382B1 (en) | Method for processing images and the corresponding electronic device | |
| KR20010043395A (en) | Method and apparatus for reducing breathing artifacts in compressed video | |
| CN1215962A (en) | Picture signal processing method and apparatus | |
| JP2007306619A (en) | Computer execution method and recording medium for processing video image | |
| JP2004032718A (en) | System and method for processing video frame by fading estimation/compensation | |
| JPWO2007043609A1 (en) | Image encoding method, and apparatus and computer program using the same | |
| CN114009015A (en) | Transmission encoding technique for converting omitted blocks | |
| JP5746193B2 (en) | Efficient adaptive filtering method and apparatus for video encoding and decoding | |
| CN102057679B (en) | Compressed image noise removal device and reproduction device | |
| EP2014099A2 (en) | Method and apparatus for encoding/transcoding and decoding | |
| JP2006067302A (en) | Apparatus, method, and program for image encoding | |
| KR20130049736A (en) | Method and apparatus for inter prediction | |
| US7161984B2 (en) | Video reproduction apparatus with fast post-processing and method thereof | |
| WO2005038719A2 (en) | Method and apparatus for calculating an inverse dct | |
| KR20180107067A (en) | Apparatus for image coding/decoding and the method thereof | |
| JP2004140863A (en) | Motion picture decoding apparatus | |
| US20200413105A1 (en) | Metric driven adaptive processing for video encoders | |
| Gowripushpa et al. | Implementation of ROI based baseline sequential adaptive quantization | |
| JP2005101914A (en) | Video signal decoding apparatus | |
| JP5256095B2 (en) | Compressed image noise removal device and playback device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| ASS | Succession or assignment of patent right |
Owner name: HITACHI MAXELL LTD. Free format text: FORMER OWNER: HITACHI LTD. Effective date: 20150302 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20150302 Address after: Osaka Japan Patentee after: Hitachi Maxell, Ltd. Address before: Tokyo, Japan Patentee before: Hitachi Consumer Electronics Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20180301 Address after: Kyoto Japan Patentee after: MAXELL, Ltd. Address before: Osaka Japan Patentee before: Hitachi Maxell, Ltd. |
|
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: Kyoto Japan Patentee after: MAXELL, Ltd. Address before: Kyoto Japan Patentee before: MAXELL HOLDINGS, Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220601 Address after: Kyoto Japan Patentee after: MAXELL HOLDINGS, Ltd. Address before: Kyoto Japan Patentee before: MAXELL, Ltd. |